{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "https://spectrai-initiative.github.io/bio-benchmark-atlas/schema/nucleic-acid-results.schema.json",
  "title": "BioBench Atlas nucleic-acid benchmark results artifacts",
  "oneOf": [
    { "$ref": "#/$defs/latest" },
    { "$ref": "#/$defs/manifest" },
    { "$ref": "#/$defs/catalog" },
    { "$ref": "#/$defs/entities" },
    { "$ref": "#/$defs/evidence" },
    { "$ref": "#/$defs/usageIndex" },
    { "$ref": "#/$defs/protocolChunk" }
  ],
  "$defs": {
    "sha256": {
      "type": "string",
      "pattern": "^[0-9a-f]{64}$"
    },
    "row": {
      "type": "object",
      "additionalProperties": { "type": ["string", "number", "boolean", "null", "array", "object"] }
    },
    "asset": {
      "type": "object",
      "additionalProperties": false,
      "required": ["path", "sha256", "rowCount", "compressedBytes", "uncompressedBytes"],
      "properties": {
        "path": { "type": "string", "pattern": "^[^/].*\\.json\\.gz$" },
        "sha256": { "$ref": "#/$defs/sha256" },
        "rowCount": { "type": "integer", "minimum": 0 },
        "compressedBytes": { "type": "integer", "minimum": 1 },
        "uncompressedBytes": { "type": "integer", "minimum": 1 }
      }
    },
    "common": {
      "type": "object",
      "required": ["schema_version", "snapshot_date"],
      "properties": {
        "schema_version": { "const": "nucleic-acid-results-v1" },
        "snapshot_date": { "type": "string", "format": "date" }
      }
    },
    "latest": {
      "allOf": [
        { "$ref": "#/$defs/common" },
        {
          "type": "object",
          "additionalProperties": false,
          "required": ["schema_version", "snapshot_date", "manifest_path", "available_snapshots"],
          "properties": {
            "schema_version": true,
            "snapshot_date": true,
            "manifest_path": { "type": "string", "pattern": "^[0-9]{4}-[0-9]{2}-[0-9]{2}/manifest\\.json$" },
            "available_snapshots": {
              "type": "array",
              "minItems": 1,
              "uniqueItems": true,
              "items": { "type": "string", "format": "date" }
            }
          }
        }
      ]
    },
    "manifest": {
      "allOf": [
        { "$ref": "#/$defs/common" },
        {
          "type": "object",
          "additionalProperties": false,
          "required": ["schema_version", "snapshot_date", "literature_cutoff", "counts", "assets", "protocolChunks"],
          "properties": {
            "schema_version": true,
            "snapshot_date": true,
            "literature_cutoff": { "type": "string", "format": "date" },
            "counts": {
              "type": "object",
              "additionalProperties": { "type": "integer", "minimum": 0 }
            },
            "assets": {
              "type": "object",
              "additionalProperties": false,
              "required": ["catalog", "entities", "evidence", "usage_index"],
              "properties": {
                "catalog": { "$ref": "#/$defs/asset" },
                "entities": { "$ref": "#/$defs/asset" },
                "evidence": { "$ref": "#/$defs/asset" },
                "usage_index": { "$ref": "#/$defs/asset" }
              }
            },
            "protocolChunks": {
              "type": "object",
              "minProperties": 1,
              "additionalProperties": { "$ref": "#/$defs/asset" }
            }
          }
        }
      ]
    },
    "catalog": {
      "allOf": [
        { "$ref": "#/$defs/common" },
        {
          "type": "object",
          "additionalProperties": false,
          "required": ["schema_version", "snapshot_date", "literature_cutoff", "benchmarks", "tasks", "tracks", "task_benchmark", "coverage", "protocols", "metrics", "summaries", "leaders", "benchmark_crosswalk", "task_crosswalk"],
          "properties": {
            "schema_version": true,
            "snapshot_date": true,
            "literature_cutoff": { "type": "string", "format": "date" },
            "benchmarks": { "type": "array", "items": { "$ref": "#/$defs/row" } },
            "tasks": { "type": "array", "items": { "$ref": "#/$defs/row" } },
            "tracks": { "type": "array", "items": { "$ref": "#/$defs/row" } },
            "task_benchmark": { "type": "array", "items": { "$ref": "#/$defs/row" } },
            "coverage": { "type": "array", "items": { "$ref": "#/$defs/row" } },
            "protocols": { "type": "array", "items": { "$ref": "#/$defs/row" } },
            "metrics": { "type": "array", "items": { "$ref": "#/$defs/row" } },
            "summaries": { "type": "array", "items": { "$ref": "#/$defs/row" } },
            "leaders": { "type": "array", "items": { "$ref": "#/$defs/row" } },
            "benchmark_crosswalk": { "type": "array", "items": { "$ref": "#/$defs/row" } },
            "task_crosswalk": { "type": "array", "items": { "$ref": "#/$defs/row" } },
            "track_coverage": { "type": "array", "items": { "$ref": "#/$defs/row" } }
          }
        }
      ]
    },
    "entities": {
      "allOf": [
        { "$ref": "#/$defs/common" },
        {
          "type": "object",
          "additionalProperties": false,
          "required": ["schema_version", "snapshot_date", "participants", "configurations", "works"],
          "properties": {
            "schema_version": true,
            "snapshot_date": true,
            "participants": { "type": "array", "items": { "$ref": "#/$defs/row" } },
            "configurations": { "type": "array", "items": { "$ref": "#/$defs/row" } },
            "works": { "type": "array", "items": { "$ref": "#/$defs/row" } }
          }
        }
      ]
    },
    "evidence": {
      "allOf": [
        { "$ref": "#/$defs/common" },
        {
          "type": "object",
          "additionalProperties": false,
          "required": ["schema_version", "snapshot_date", "sources", "result_sources", "leaderboard_snapshots"],
          "properties": {
            "schema_version": true,
            "snapshot_date": true,
            "sources": { "type": "array", "items": { "$ref": "#/$defs/row" } },
            "result_sources": { "type": "array", "items": { "$ref": "#/$defs/row" } },
            "leaderboard_snapshots": { "type": "array", "items": { "$ref": "#/$defs/row" } }
          }
        }
      ]
    },
    "usageIndex": {
      "allOf": [
        { "$ref": "#/$defs/common" },
        {
          "type": "object",
          "additionalProperties": false,
          "required": ["schema_version", "snapshot_date", "benchmark_protocols", "task_protocols", "track_protocols", "participant_protocols", "configuration_protocols", "work_protocols", "protocol_result_counts", "benchmark_result_counts", "participant_result_counts"],
          "properties": {
            "schema_version": true,
            "snapshot_date": true,
            "benchmark_protocols": { "$ref": "#/$defs/stringListMap" },
            "task_protocols": { "$ref": "#/$defs/stringListMap" },
            "track_protocols": { "$ref": "#/$defs/stringListMap" },
            "participant_protocols": { "$ref": "#/$defs/stringListMap" },
            "configuration_protocols": { "$ref": "#/$defs/stringListMap" },
            "work_protocols": { "$ref": "#/$defs/stringListMap" },
            "protocol_result_counts": { "$ref": "#/$defs/countMap" },
            "benchmark_result_counts": { "$ref": "#/$defs/countMap" },
            "participant_result_counts": { "$ref": "#/$defs/countMap" }
          }
        }
      ]
    },
    "stringListMap": {
      "type": "object",
      "additionalProperties": {
        "type": "array",
        "uniqueItems": true,
        "items": { "type": "string", "minLength": 1 }
      }
    },
    "countMap": {
      "type": "object",
      "additionalProperties": { "type": "integer", "minimum": 0 }
    },
    "protocolChunk": {
      "allOf": [
        { "$ref": "#/$defs/common" },
        {
          "type": "object",
          "additionalProperties": false,
          "required": ["schema_version", "snapshot_date", "protocol_id", "protocol_fingerprint", "benchmark_id", "task_id", "track_id", "results"],
          "properties": {
            "schema_version": true,
            "snapshot_date": true,
            "protocol_id": { "type": "string", "minLength": 1 },
            "protocol_fingerprint": { "type": "string", "pattern": "^PF-v1-" },
            "benchmark_id": { "type": "string", "pattern": "^B[0-9]{2}$" },
            "task_id": { "type": "string", "pattern": "^T[0-9]{2}$" },
            "track_id": { "type": "string" },
            "results": {
              "type": "array",
              "items": {
                "allOf": [
                  { "$ref": "#/$defs/row" },
                  {
                    "type": "object",
                    "required": ["result_id", "protocol_id", "metric_id", "participant_id", "configuration_id", "work_id", "canonical_value", "canonical_evidence_source_id", "sources"],
                    "properties": {
                      "sources": {
                        "type": "array",
                        "minItems": 1,
                        "items": { "$ref": "#/$defs/row" }
                      }
                    }
                  }
                ]
              }
            }
          }
        }
      ]
    }
  }
}